A5103022313- Machine Learning in Materials Science
- Advanced Image Processing Techniques
- Protein Structure and Dynamics
- Sentiment Analysis and Opinion Mining
- Spectroscopy and Quantum Chemical Studies
- Granular flow and fluidized beds
- Advanced Chemical Physics Studies
- Topic Modeling
- Advanced Image Fusion Techniques
- Caching and Content Delivery
- Music and Audio Processing
- Speech Recognition and Synthesis
- Image Processing Techniques and Applications
- Organic Light-Emitting Diodes Research
- IoT and Edge/Fog Computing
- Photochromic and Fluorescence Chemistry
- Image and Signal Denoising Methods
- Multimodal Machine Learning Applications
- Forest Biomass Utilization and Management
- Advanced Optical Sensing Technologies
- Advanced Text Analysis Techniques
- Digital Imaging for Blood Diseases
- Target Tracking and Data Fusion in Sensor Networks
- Semiconductor Quantum Structures and Devices
- Perovskite Materials and Applications
Huaqiao University
2022-2024
Shaoxing University
2024
Beijing Normal University - Hong Kong Baptist University United International College
2024
Beijing Normal University
2024
Hong Kong Baptist University
2024
Xidian University
2024
Xi'an Polytechnic University
2024
China Jiliang University
2024
Wuhan University of Technology
2021-2023
Wuchang University of Technology
2023
In the simulation of molecular systems, underlying force field (FF) model plays an extremely important role in determining reliability simulation. However, quality state-of-the-art fields is still unsatisfactory many cases, and FF parameterization process largely relies on human experience, which not scalable. To address this issue, we introduce DMFF, open-source development platform based automatic differentiation technique. DMFF serves as a powerful tool for both top-down bottom-up...
We employed RbI additive, constructed heterojunction, and used TOPO post-treatment for suppressing non-radiative recombination of MA-free WBG perovskite. The device showed a record PCE 23.35%, high V OC 1.3 the impressive stability.
Singing voice synthesis (SVS) systems are built to synthesize high-quality and expressive singing voice, in which the acoustic model generates features (e.g., mel-spectrogram) given a music score. Previous models adopt simple loss L1 L2) or generative adversarial network (GAN) reconstruct features, while they suffer from over-smoothing unstable training issues respectively, hinder naturalness of synthesized singing. In this work, we propose DiffSinger, an for SVS based on diffusion...
Although the image recognition has been a research topic for many years, researchers still have keen interest in it[1]. In some papers[2][3][4], however, there is tendency to compare models only on one or two datasets, either because of time restraints model tailored specific task. Accordingly, it hard understand how well certain generalizes across field[6]. this paper, we four neural networks MNIST dataset[5] with different division. Among them, three are Convolutional Neural Networks...
An accurate, transferrable, and computationally efficient potential energy surface is of paramount importance for all molecular mechanics simulations. In this work, by using water as an example, we demonstrate how one can construct a reliable force field combining the advantages both physically motivated data-driven machine learning methods. Different from existing models based on many-body expansion, adopt separation scheme that completely distances, which more convenient generic systems....
In the simulation of molecular systems, underlying force field (FF) model plays an extremely important role, determining reliability simulation. However, quality state-of-the-art fields is still unsatisfactory in many cases, and FF parameterization process largely relies on human experience, which not scalable. To address this issue, we introduce DMFF, open-source development platform based automatic differentiation technique. DMFF serves as a powerful tool for both top-down bottom-up...
Recently, deep-learning-based image super-resolution methods have made remarkable progress. However, most of these do not fully exploit the structural feature input image, as well intermediate features from layers, which hinders ability detail recovery. To deal with this issue, we propose a gradient-guided and multi-scale network for (GFSR). Specifically, dual-branch structure is proposed, including trunk branch gradient one, where latter used to extract map prior guide reconstruction...
Accurate and efficient species classification is crucial for various applications, machine-learning approaches have been widely employed image-based taxonomic identification. However, challenges arise due to the variability in image quality need extensive dataset preparation. In this study, we propose a model that combines unsupervised semantic segmentation by distilling feature correspondences (STEGO) no-reference assessment assess of images STEGO enables counting objects an image, while...
In this paper, an X-ray computed tomography (CT) imaging-based and machine learning-enabled framework is presented to characterize multi-constituent granular materials. framework, CT first utilized obtain raw images of a synthetic material. Then, learning tool termed trainable Weka segmentation (TWS) implemented segment images, i.e., classify constituents segregate particles in contact. This fundamentally different approach for image that it predicts results based on trained classifier model...
<b><sc>Abstract.</sc></b> Discrete element method (DEM)-based particulate flow simulations of wedge hopper discharge milled loblolly pine are performed using an experiment-validated nonlinear hysteretic interparticle contact model. The DEM simulation-based study enables detailed and quantitative analyses the clogging behavior based on fundamental particle physics. Wide ranges critical processing parameters (CPPs) (of hopper) material attributes (CMAs) pine) considered in analysis. Among...
The diverse applications of chlorinated paraffins (CPs) has given rise to widespread environmental diffusion, close attention paid CPs due it can bioaccumulate in lipid tissues organisms and be transferred magnified along the food chain. Among them, short-chain (SCCPs, C10–C13) is classified as persistent organic pollutants (POPs) under Stockholm Convention. As known, molecular structure determines their physicochemical properties toxicities, so, strcuture characterization technical products...